Your ads promise one thing. Your landing page says another. I built https://teapawt.org/t.co/DIzW0eAmYr to catch that - automatically. | Meta/Google App Approved |
AD ↔️ Landingpage Alignment ➡️
Joined August 2009
- Tweets4.8K
- Following566
- Followers1.3K
- Likes4.1K
Pinned Tweet
Day 22 - Roasting my way to 1,000,000 ad-to-landing page alignment evaluations | Currently 402 teapawt.org/i/broadcasts/1nJOLLVEE…
Your ad promises 50% off. The visitor clicks.
The landing page says nothing about 50% off.
Now they doubt the offer was ever real.
So they leave. Your CPA climbs.
You blame the creative. The creative was fine. The page broke the promise.
Tried to roast AG1 today. Got a 1/10.
Not because the alignment is bad. Because their anti-bot security blocked the screenshot.
That's a tooling score, not a real one. I won't pretend a blocked scrape is a teardown.
Bigger brands, bigger walls. Fixing the fallback tomorrow.
Day 22 of scoring ads against landing pages on stream.
I call my score out loud before the engine runs.
Today I nailed one (3.9, called a 4). Missed one badly (called a 6, scored 3.4).
My bias: I credit brands for "the product's on the page somewhere." The engine doesn't.
MVMT ad: watches, 50% off.
Their landing page: no 50% off. No promo code. A free shipping banner that was never in the ad. Even a different watch.
Message match is the dimension that matters most on Meta. This one fails it.
Predicted 5.7. Scored 4.5.
Two brands today ran paid ads built around a specific offer.
Both sent the click to their homepage.
Danner scored 3.4. ClassPass dumped a "two weeks of Pilates" ad onto its home page.
Homepage dumping is the most common way I watch ad budgets leak. No platform flags it.
I'm not selling my product. I'm roasting bad ads in public — and it's building the audience AND the product at the same time.
Danner ran a "50% off your first purchase, sign up with email" ad straight to their homepage.
No discount on the page. No email signup. The offer just disappears.
I predicted a 6. AdAlign scored it 3.4.
I was being generous. The engine wasn't.
Called a 4 before running the engine. It scored 3.9. Closest prediction I've made.
Ten years reading ads and I'm finally calibrated to my own tool. Reassuring or concerning, not sure yet.
True Protein: 5.5/10 on ad-to-page alignment.
Tone alignment 8. Visual match 4.
The page keeps the promise. It just doesn't look the same as the ad.
Right message. Wrong visual. Its the gap is where paid traffic leaks.
Scored a Reddit ad against its landing page: 5/10.
Message matched (7).
But the page looked like a different company (visual 3) and read in a different voice (tone 4.5).
Right words, wrong face.
On Reddit the vibe IS the message.
Building a box on the AdAlign homepage: paste your ad, get an alignment score against your landing page. Free.
The open question is gating it behind an email.
I don't want to spend my energy fighting abuse and IP scrapers.
Where would you draw that line?
I call the alignment score out loud before I run the engine.
Today I said 5. The engine said 4.
When my gut and the engine land in the same band, that's the signal the scoring is real.
The day they stop agreeing is the day I stop trusting the number.
Scored an Amazon ad against its landing page today.
The ad points at a specific product. The page it lands on is the Amazon homepage.
That's not a landing page. That's a parking lot.
4/10. The page never finishes the sentence the ad started.
Day 21.
Two advertisers connected their ad accounts this week.
Now most of the ad-to-page pairs I'm scoring come from those, not from me hunting ads manually.
The manual roasts were always the bootstrap. The real dataset comes from accounts that never stop running.
Advice I got this week: gate the free tier.
My instinct was no. Free should be frictionless.
But an open free endpoint is an open door for abuse. One email = one cap = a much smaller attack surface.
Day 21 - Roasting my way to 1,000,000 ad-to-landing page alignment evaluations | Currently 397 teapawt.org/i/broadcasts/1YGNrrELo…